ci: stop the Dependabot docker job that has failed every week for two months - #43
Merged
Conversation
… months
The "docker in /. - Update" run on main has failed on eight consecutive Mondays
(2026-06-07 through 2026-07-26) and never once opened a pull request.
Cause, from the run log:
GET https://public.ecr.aws/v2/docker/library/golang/tags/list
401 {"errors":[{"code":"DENIED","message":"Not Authorized"}]}
The Dockerfiles pin base images on ECR Public to dodge Docker Hub pull rate
limits in CI, and Dependabot's docker updater cannot authenticate against ECR
Public to list tags. The job could therefore never do its job — it was pure red
noise on the default branch, which is worse than no signal because it trains
everyone to ignore a failing check.
api and ui hit exactly this and removed the ecosystem two months ago (api#213,
ui#194). agent was missed, so it kept failing.
Base-image CVEs remain covered: the "Docker Image Scan" job in
.github/workflows/security.yml builds the image and runs Trivy over it at
CRITICAL,HIGH (verified running, not skipped). That job builds the default
Dockerfile only, so the variant images rely on the filesystem Trivy scan plus
manual bumps — noted in the config comment.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The bug
docker in /. - Updatehas failed on eight consecutive Mondays onmain—2026-06-07, 06-14, 06-21, 06-28, 07-05, 07-12, 07-19, 07-26 — and has never once
opened a pull request.
From the run log:
The Dockerfiles pin base images on ECR Public to dodge Docker Hub pull rate limits
in CI, and Dependabot's docker updater cannot authenticate against ECR Public to
list tags. The job could never do its job.
That makes it worse than no check at all: a permanently red run on the default
branch teaches everyone to ignore a failing check.
Why agent and not api/ui
api and ui hit the identical failure and removed the ecosystem two months ago
(api#213, ui#194). agent was missed — so it kept failing every week
while the other two went quiet. This closes the gap.
Found by taking the latest run of every workflow on each repo's shipping branch;
agent's docker job was the only thing still red anywhere across api / ui / agent /
sdk-go / ctis.
This does not lose CVE coverage
Nothing is lost by removing this, because the Dependabot docker job never
produced anything — but it is worth being precise about what does cover base
images, and where the edges are.
The
Docker Image Scanjob in.github/workflows/security.ymlbuilds the imageand runs Trivy over it at
CRITICAL,HIGH. It genuinely runs: the last Securityrun on
mainreports success for it, alongside Trivy FS, CodeQL, LicenseCompliance and Go Vulnerability Check. (It shows as skipping on this PR — see
the gate below.)
Two edges I would rather state than gloss over:
main(
if: github.event_name == 'push' && github.ref == 'refs/heads/main'), so itdoes not run on pull requests or on the weekly schedule. A CVE newly
disclosed in a base image is therefore not surfaced by the image scan until the
next push to
main. The weekly scheduled run still covers Trivy FS, CodeQL andGo Vulnerability Check — just not the built image.
Dockerfileonly, so the four variant images(nuclei / gitleaks / semgrep / trivy) get the filesystem scan plus manual
bumps, not an image scan.
Neither is a regression from this PR — both are true today. Dropping the
schedulegate ondocker-scanand adding the variants would close them, and isa separate change.
Verification
.github/dependabot.ymlparses as valid YAML; remaining ecosystems aregomodandgithub-actions, both unchanged and both currently green.public.ecr.awsexcept the distroless runtime stage in
Dockerfile, which is ongcr.io. Thecomment says so rather than overstating it.